Evaluation of Cruise Ship Supply Logistics Service Providers with ANP-RBF

Table 1

Summary of LSPs evaluation methods.

Method	Advantages	Shortcomings
AHP	It can simplify complex problems; there are few requirements for quantitative data information.	It is too subjective, and too many indexes will lead to large data statistics and uncertain weight. With the increase of indexes, the accurate calculation of eigenvalues and eigenvectors will be more complex.
DEA	It selects high quality suppliers by measuring the relative efficiency of evaluation objects, and it is simple to apply.	It is usually a subjective value determined in advance and specific to a certain degree of randomness.
TOPSIS	It requires less sample data and is easy to understand and simple to operate.	The distance as a scale can only reflect the position relationship between data curves and cannot reflect the dynamic changes of data series.
BP neural network	It can solve any complex nonlinear problems and has the ability to self-learn and self-adapt.	It has low learning efficiency and slow convergence speed, is easy to fall into local minimum state, and so on.
Fuzzy comprehensive evaluation	It is simple and easy to understand, and the evaluation effect is ideal for complex problems with multiple indexes and multiple levels.	The calculation of the weight of each index is highly subjective, and it is difficult to determine the membership function.
Grey relational analysis	The grey information and white information are treated differently, which can ensure the integrity and accuracy of information; they do not need many samples and normalization processing.	The theoretical basis is relatively narrow, the standardization lacks comprehensiveness, and it is unable to deal with the information duplication caused by relevant indicators.
ANP	An interactive risk evaluation system is incorporated into the supplier selection process. The triangular fuzzy number is used to construct the linguistic evaluation set to avoid the negative impact of subjective judgment.	It is a very complex operation process to solve the hypermatrix problem.